Repository navigation
Conversation
Fold constant subgraphs, including cat/split chains, before ATen decomposition so linear and addmm biases remain eligible for delegation. Skip folding when a mutable operator could write into a folded constant. Serialize only the bytes belonging to each prepared tensor rather than its backing storage. Prepared XNNPACK constants may be views, so using storage pointers and sizes ignored storage offsets and included unrelated bytes. Reading from data_ptr() for numel() * element_size() bytes gives split views, parameter views, and quantization scales the correct payload, size, and named-data hash. For the static Roblox Voice Safety V2 model, median execution latency improves from about 16 s to about 0.5 s: roughly 35x faster. Signed-off-by: Måns Nilsson <mans.nilsson@arm.com> Change-Id: I988ec6f77d3d7686ffc8a2ad09e5675948a6d4be
mansnils
added
the
module: xnnpack
mansnils
added
partner: arm
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/23154
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New FailureAs of commit fdcbee1 with merge base 7586c16 ( NEW FAILURE - The following job has failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
Contributor
The Python binding borrows the PTE bytes. Retain each buffer until execution finishes so XNNPACK can read named constants reliably. Signed-off-by: Måns Nilsson <mans.nilsson@arm.com> Change-Id: I91b173ac511ff62f5d93b6faf19c42b0956e5e42
Collaborator
Author
# Conflicts: # backends/xnnpack/operators/node_visitor.py
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fold constant subgraphs, including cat/split chains, before ATen decomposition so linear and addmm biases remain eligible for delegation. Skip folding when a mutable operator could write into a folded constant.
Serialize only the bytes belonging to each prepared tensor rather than its backing storage.
Prepared XNNPACK constants may be views, so using storage pointers and sizes ignored storage offsets and included unrelated bytes. Reading from data_ptr() for numel() * element_size() bytes gives split views, parameter views, and quantization scales the correct payload, size, and named-data hash.
For the static Roblox Voice Safety V2 model, median execution latency improves from about 16 s to about 0.5 s: roughly 35x faster.
cc @GregoryComer @digantdesai @cbilgin @JakeStevens @freddan80 @per @zingo @oscarandersson8218 @Sebastian-Larsson @robell @rascani